Following Trends
motivated by e-mail from Salvatore DeM.

There's an interesting article about a Buy & Sell ritual that ...

>Not another one!
Pay attention. We're going to do the following:

  • We download ten years worth of weekly stock prices.
  • At the end of each week we look at the maximum of Friday's Closing stock Price, from ten years ago until this Friday's Close.
    We're talking about an "all time high", from the starting week.
  • If Friday's Close is at that Maximum, we assume the stock is on an uptrend ... and we BUY at Monday's Open..
  • At the end of each week we also calculate the True Range, averaged over the past 8 weeks. ... about 40 market days.
    The True Range (TR) identifies the largest price variation from last Friday's Close to this Friday's Close.
  • We compare this Friday's Close to 10x the ATR value.
  • If it's less than (or equal to) 10ATR, we SELL at the Open, on the following Monday.

TR : The Maximum Variation in Price

>Sell when the Price falls below 10xATR? That seems curious. I mean ...
Think of it this way:   ATR (averaged over the last 8 weeks), is a fraction f of the current Price P, namely f = ATR / P ... so ATR = f P.
Example: f = ATR/P = 0.08, meaning the maximum price variation is 8% of the current Price.
Then 10 ATR = 10 (f P) and that's when we Sell ... when the Price drops below that.
Example: If f = ATR/P = 0.08, we sell when the Price drops below 10 f P = 0.80 P. That's a drop of 20%.

For example, in the picture of the spreadsheet (below), there's a Sell signal on Friday, Oct 2, 1998.
At that time, the Price was just $19.79 and ATR was $2.112.
We Sell because the Price had dropped below 10 ATR = $21.12.
We sell at Monday's Open (which happened to be $19.62.)
The week before, P = $21.74 and 10 ATR = $17.30 ... so no Sell signal was generated.

>That's it?
Yes. We have a BUY signal, based upon the running maximum ... and a SELL signal based upon the Average True Range. Now we just ...

>Does it work good?
That's what I was going to talk about.
We got a spreadsheet, like so:   (Click on the picture to download.)

After downloading, we can pick the number of weeks to average (Example: 8 weeks) and the multiplier (Example: 10).
We start with umpteen Shares and/or $Cash in our portfolio ... then follow the trend as described above. (In the sample spreadsheet, intial Shares were 0 and initial Cash was $100K.

>But you get 12% Compound Annual Growth Rate instead of 11%? That's good?
But it'll depend upon the stock and the time period and what you choose for the number of weeks in the average and the ATR multiplier and the phases of the moon etc. etc.
Here are a few more examples:

Note that, in this lousy market environment, you wind up selling everything and sitting on Cash.
Of course, if the asset has a low volatility, you never sell. For example, here's a Bond Fund:

>So what about some other time period?
Do this:
[1] Download the spreadsheet.
[2] Pick ten years worth of time.
[3] Click a button.
[4] Change what you find in the red boxes, the initial Shares and/or Cash holdings.
[5] Play ...

You'll (sometimes) find that, for certain time periods and parameter choices, you Sell and miss the subsequent rise in price 'cause it don't hardly reach the moving maximum ... after a big drop.

>So why don't you change the parameters, like that 10 or maybe the 8 ...
I already told you! Do this:
[1] Download the spreadsheet.
[2] Pick ten years worth of time.
[3] Click a button.
[4] Change what you find in the red boxes, the initial Shares and/or Cash holdings.
[5] Play ... and imagine a future just like the past.


a Modification

If you had used the above scheme during the crash of 1929, you'd have sold ... then stayed in Cash for some 20 or 30 years.

Suppose we only look back at the Max Price for, say, 50 weeks.
Then we'd get something like this (for our earlier example of the GE):

Now you won't miss those big increases after the big drop and ...

>So you really believe in this stuff?
Huh? Of course! My computer says so and I always believe my computer.
If you can't believe your computer, who can you believe?

>I have this bridge I'd like to sell you. It's in very good shape and ...




I forgot to mention that there's a cell in the spreadsheet that says 2000.
It means "calculate the Max over the past 2000 weeks" ... which really means from the initial time.
If you leave it at 2000 you get the Buy/Sell ritual we've been explaining.
BUT, it you change it to, say, 50, you BUY when the price is at the Max over the past 50 weeks.

>I don't see that 2000 in the picture of the spreadsheet.

Another thing:
Since we've modified the BUY signal (so we only look at the Max over the past umpteen weeks rather than over all weeks), we should ...

>Modify the SELL signal as well.
You took the words right outta my mouth.
I reckon there are lots of ways to do that, but let's so something familar: the Lower Bollinger Band.
Recall that we calculate the Average Price over the past umpteen weeks as well as the Standard Deviation of Prices over this same time period.
Then the Lower Bolli is: (Average Price) - k (Standard Deviation) where k is some number ... like maybe 2 or 3.
So, when the Price drops to 2 Standard Deviations below the Moving Average Price, we figure it's dropped enough ... so we SELL.

>I assume the spreadsheet has that feature? I mean, I can choose either the ATR SELL signal or the Bolli?
Yes ... and more!